Cooling Tower Fans and Gearboxes: The Rotating Equipment You Cannot Walk Up To

• The drive shaft, gearbox and fan sit inside the tower's fan stack in a wet, hot, humid space. Reaching them means lockout, fall protection and a climb, so they are often watched least.
• Cutout switches catch gross imbalance but typically miss the higher frequency vibration of gear defects, bearing faults and poor lubrication. They give no warning and no data.
• An Electric Power Research Institute (EPRI) field evaluation tied frequent fan starts and stops to gearbox failures. A new gearbox runs about $30,000 plus about $5,000 in labor.
• Freeze protection is a procedure, not a piece of equipment. The CME outage shows how a tower draining step skipped in cold weather can become a whole plant failure.

A typical induced draft tower fan drive has four parts: a motor, a long drive shaft, a right angle gearbox and a large fan. The fan sits at the top and pulls air up through the tower.

• The motor sits outside the fan stack, where it stays drier and easier to reach.
• The drive shaft, also called a jackshaft, runs from the motor to the gearbox inside the fan cylinder.
• The right angle gearbox turns the horizontal shaft drive into vertical fan rotation and reduces speed.
• The fan is large in diameter and turns slowly, moving a great deal of air with each revolution.

Every link in that chain is rotating equipment, with the same faults covered earlier in this series. The motor can overheat as in Part 6, and the shaft can be misaligned as in Part 2. The gearbox has bearings that fail as in Part 1.

What makes the tower different is where those parts live. The drive train spends its whole life in warm, saturated air and spray.

Tower fans and gearboxes are hard to inspect because they sit inside the tower's shroud, in a wet, hot, humid space. Getting to them means lockout, fall protection and climbing onto the fan deck.

That makes every close inspection a planned job rather than a glance on a round. It is the main reason the gearbox and fan are often the least watched rotating equipment on site.

The result is a gap in coverage. A pump in the mechanical room gets looked at every day, while the gearbox above the tower may go months between visits.

Multi cell towers help close that gap. Cells built alike and running alike should sound, look and run at similar temperatures. The cell that differs stands out from the ground or the deck edge.

Much of what matters can still be checked from outside the stack. Sound, oil streaks, motor temperature, fan speed and the oil sight glass piped to the deck all qualify. Those checks are the core of this article.

Cooling tower gearboxes fail mainly through lubrication problems, gear tooth wear, bearing faults and oil leaks. Lubrication comes first, because the tower environment works against it.

• Insufficient or contaminated lubrication: water ingress from the humid tower air is common, and wet oil protects gears and bearings poorly.
• Gear tooth wear: surfaces pit and wear, especially when lubrication is poor or loads change often.
• Bearing faults: the gearbox's bearings fail in the same stages described in Part 1.
• Oil leaks: seals wear, oil level drops, and the gearbox runs short of lubricant.

A failed gearbox can also contaminate the tower water with oil. That turns a mechanical repair into a water treatment problem as well.

Why water in the oil matters

Warm, saturated air surrounds the gearbox all day. As the gearbox heats and cools, it breathes that air in through its vent, and moisture condenses inside.

Water thins the oil film between gear teeth and promotes rust on bearing surfaces. Oil that looks milky or cloudy in a sight glass is a sign worth recording, and an oil sample confirms it.

What gear mesh frequency tells you

Gear mesh frequency equals the number of teeth times the shaft speed. It is the rate at which teeth engage, and it shows as a distinct peak in a vibration spectrum.

Sidebands spaced at shaft speed around that peak point to gear wear or a damaged tooth. Reading them takes vibration analysis with access to the fan deck, not a walk past.

Checking the oil

Oil level is often checked through an external sight glass or fill line at the fan deck. Oil streaks on the fan cylinder or deck point to a leak.

Cooling tower fan imbalance is caused by uneven mass or uneven aerodynamic load across the blades. Because the fan is large and slow, imbalance shows at low frequency, once per fan revolution.

The common causes are:

• Blade pitch that differs from blade to blade, so some blades push more air than others.
• Water absorbed into hollow blades, adding weight to some blades and not others.
• Ice building unevenly on blades in cold weather.
• A broken or missing blade tip.
• Loose hub clamps, letting a blade shift or change pitch.

Imbalance shakes the whole stack and loads the gearbox and shaft bearings. The general signature of imbalance is covered in Part 2; on a tower, visible stack shaking is often the first sign.

A thrown blade is the extreme case. It can wreck the fan cylinder and the deck around it.

A vibration cutout switch protects against gross shaking, such as a thrown blade or severe imbalance, by tripping the fan. It misses most of the faults that develop slowly inside the gearbox.

Vibration switches are the oldest and most basic protection on tower fans. Gear defects, bearing faults and poor lubrication produce higher frequency vibration that these switches typically do not respond to (Emerson).

Why a switch is not monitoring

A switch gives no advance notice and no data. It is either tripped or not, and it says nothing about the weeks of wear that came before.

Because switches are rarely monitored, the first sign that one has stopped working may be a fan running to failure. Corrosion in the tower environment can also disable them.

That is why a switch should be tested on a schedule, and its status checked to confirm it has not been bypassed. A bypassed switch removes the last line of protection without anyone noticing.

A continuous vibration record does what a switch cannot. It shows a gear or bearing fault growing over weeks, so the repair can be planned for a mild day instead of forced on the hottest one.

Yes. An EPRI field evaluation, reported in Combined Cycle Journal, found that frequent fan starts and stops lead to gearbox failures (Combined Cycle Journal).

The test saw four gearbox failures in one year. Each start loads the gear teeth hard as the large fan accelerates from rest.

What a failure costs

A new gearbox runs about $30,000, plus about $5,000 in labor. Four failures in a year put the cost of on and off cycling in plain terms.

What EPRI recommended

• Continuous gearbox temperature monitoring.
• Real time vibration monitoring.
• Better airflow around the gearbox.

All three replace an occasional climb with a steady record. They close the gap that the tower's location creates.

Variable frequency drives (VFDs) are generally good for tower fans. In the EPRI evaluation, VFD speed control gave the greatest net benefit compared with on and off control and two speed control.

A VFD lets the fan ramp gently and run continuously at the speed the load needs. That removes most of the hard starts that damage gears.

Limits to respect

Gearboxes are generally designed for one way rotation. Reversing fans for de icing, and frequent speed swings, add wear.

Check the manufacturer's minimum speed and reversing limits before setting up the drive. Running below the minimum can starve some gearboxes of lubrication, and some speeds can excite resonance, covered in Part 5.

In freezing weather, standing water in a cooling tower can freeze in basins, piping and fill, and ice builds on fans and louvers. Towers need drain down or freeze protection procedures in cold weather.

Ice also adds uneven weight to fan blades, which causes imbalance. Reversing a fan to clear ice brings the gearbox wear described above.

What happened at CME and CyrusOne

CyrusOne attributed the November 2025 outage to procedures for draining cooling towers not being followed (DatacenterDynamics). CME said initial remediation attempts made the problem worse, and a number of chillers then failed.

Temperatures inside rose above 100°F (38°C), and CME's markets were halted for more than 10 hours. CyrusOne said it added cooling redundancy afterward (Insurance Journal).

The lesson for any plant

Freeze protection is a written procedure with a named owner. The tower problem became a chiller problem, and then a whole plant problem.

The procedure is only as good as its execution on the night it is needed. Who drains which cell, in what order, and who confirms it, all need to be written down before the cold arrives.

The same cold weather discipline applies to the rest of a building, covered in our hard freeze checklist.

Direct drive is a growing design trend for cooling tower fans. Permanent magnet motors that drive the fan directly remove the drive shaft and gearbox entirely.

Removing the gearbox removes its oil, its gears and its alignment to the motor. The failure modes in this article that belong to the shaft and gearbox go with them.

For an existing tower, the question usually comes up when a gearbox fails. That is the moment to compare a like for like replacement with removing the gearbox altogether.

The fan and motor still need watching. Blade imbalance, ice and motor faults remain, so the checks on sound, shaking and temperature still apply.

In both, lost heat rejection raises condenser water temperature. That raises chiller lift and surge risk, covered in Part 7.

In an office building

A lost tower cell reduces chiller capacity on the hottest days, exactly when the building needs it. A thrown fan blade can wreck the fan cylinder and deck, turning a gearbox repair into a structural one.

Tower drives belong in routine heating, ventilation and air conditioning (HVAC) maintenance, but the climb means they often slip.

In a data center

Lost tower cells reduce plant capacity. As CME showed, a tower problem can become a whole plant problem, and then an outage.

Towers often sit between teams, outside the building and outside the mechanical room. That pattern is covered in cooling loads nobody owns.

Several faults on a tower look or sound alike. Frequency and location usually separate them.

• Fan imbalance versus gearbox faults versus motor bearing faults: imbalance shows at once per fan revolution, low frequency. Gearbox faults show at gear mesh frequency, high frequency. Motor bearing faults are covered in Part 1 and Part 2.
• Drive shaft misalignment versus gearbox wear: misalignment loads the shaft couplings and shows at shaft speed, covered in Part 2. Gear wear shows as sidebands around gear mesh frequency.
• Normal splash and drift noise versus mechanical noise: falling water makes a steady broadband sound. Grinding, knocking or a rhythmic beat is mechanical.

This is why no single sensor is enough. Acoustic listening compares cells, thermal imaging compares housings, and a vibration spectrum separates fan from gear.

A walk past can prove an oil leak, low oil, a shaking stack, ice and a tripped vibration switch. It cannot prove the condition of the gears or the oil inside.

• Directly observable: oil streaks, low sight glass oil, visible fan stack shaking, ice buildup and a tripped vibration switch.
• Strong indicator: one cell sounding different from its neighbors, and a gearbox or motor running hotter than its sibling cell.
• Suspected: internal gear condition and oil quality. These need an oil sample or vibration spectrum taken with access to the fan deck.

That is the limit of any route based check. It finds the cell that differs from its neighbors; an oil sample or spectrum confirms why.

A camera records oil streaks and ice, and a walkthrough record keeps the history. Comparing cells over time is the core of predictive maintenance.

• Listen to each cell from the deck or ground, and compare cells.
• Gearbox oil sight glass level, where visible.
• Oil streaks on the fan cylinder and deck.
• Motor and gearbox housing temperatures compared across cells.
• Vibration switch status, confirmed not bypassed.
• VFD speed and on or off state for each cell.
• In cold weather, ice on blades, louvers and the fan cylinder.

Coming Soon

1. 1. When was each tower gearbox oil last changed or sampled?
2. 2. Has each vibration cutout switch been tested, and when?
3. 3. How many times a day does each tower fan start and stop?
4. 4. What is your written procedure for tower freeze protection, and who signs it off?
5. 5. Which tower cells can you see or hear without climbing onto the deck?

What does a cooling tower gearbox do?

A cooling tower gearbox is a right angle gearbox inside the fan cylinder. It takes drive from a horizontal shaft connected to the motor, turns it vertical and reduces speed to drive a large, slow turning fan.

Why do cooling tower gearboxes fail?

The main causes are insufficient or contaminated lubrication, gear tooth wear, bearing faults and oil leaks. Water ingress from the humid tower air is a common cause of contaminated oil. Frequent starts and stops also lead to gearbox failures.

How do you check cooling tower gearbox oil?

Oil level is often checked through an external sight glass or fill line at the fan deck. Oil streaks on the fan cylinder or deck point to a leak. Oil quality needs an oil sample.

What is a cooling tower vibration switch?

A vibration switch is a basic protective device that trips the fan when vibration becomes severe. It is the oldest and most basic protection on tower fans.

Is a vibration switch enough protection?

No. Gear defects, bearing faults and poor lubrication produce higher frequency vibration that switches typically do not respond to, and a switch gives no advance notice or data. Switches can also fail unnoticed through corrosion, so they should be tested.

Do frequent starts damage a cooling tower gearbox?

Yes. An EPRI field evaluation found frequent fan starts and stops lead to gearbox failures, with four failures in one year of testing. VFD speed control gave the greatest net benefit compared with on and off or two speed control.

What causes cooling tower fan imbalance?

Common causes are blade pitch that differs blade to blade, water absorbed into hollow blades, ice, a broken or missing blade tip and loose hub clamps. Imbalance shows at once per fan revolution and can shake the whole stack.

How much does a cooling tower gearbox cost?

In the EPRI evaluation reported by Combined Cycle Journal, a new gearbox ran about $30,000 plus about $5,000 in labor. Repeated failures multiply that cost quickly.

What caused the CME data center outage?

CyrusOne attributed the November 2025 outage at its Aurora, Illinois data center to procedures for draining cooling towers ahead of freezing weather not being followed. CME said early remediation made the problem worse and a number of chillers then failed. Markets were halted for more than 10 hours.

What is a direct drive cooling tower fan?

A direct drive fan uses a permanent magnet motor to turn the fan directly. It removes the drive shaft and gearbox entirely, along with their oil and alignment faults.

• Chiller Compressor Distress: Surge, Lift and the Panel Readings That Warn Before a Trip
• Imbalance, Misalignment, Looseness and Soft Foot: Four Faults, Four Vibration Signatures
• Frozen Pipes in Office Buildings: A Texas Hard Freeze Checklist
• Cooling and Mechanical: ASHRAE 180 and the Loads Nobody Owns

• Emerson: why vibration switches miss early cooling tower faults
• Combined Cycle Journal: EPRI test program on cooling tower fan reliability
• DatacenterDynamics: CME data center shutdown attributed to human error
• Insurance Journal: CME outage temperatures and added redundancy